Algolia
Algolia is a hosted search API that provides developers with tools to build fast and relevant search experiences for their applications.
Verdict
Common use cases
- Wipe staging index before bulk reindex
- Clone production index for safe testing
- Track conversion events from checkout flow
- Compare search variants with A/B tests
- Export all records for compliance audit
Integration
- Vendor
- Algolia
- Category
- developer-tools
- Auth
- API_KEY
- Tools
- 30
- Composio slug
algolia
Tools
- Add AB Test
Tool to create an ab test comparing search performance between two variants. use to test different index configurations or search parameters and measure impact on click-through and conversion rates.
- Browse Algolia Index
Tool to retrieve all records from an index. use when you need to export or iterate through an entire index dataset.
- Clear Objects
Tool to clear records of an index without affecting settings. use when you need to completely wipe all objects (e.g., after a bulk reindex) while preserving index configuration.
- Clear Rules
Tool to delete all rules in an index. use when you need to wipe all rules before re-creating them. use after confirming no critical rules require retention.
- Clicked Object IDs
Tool to send a click event to algolia to capture clicked items. use immediately after a user click outside of search contexts to report click events.
- Clicked Object IDs After Search
Tool to send a click event after a search response. use when you want to report which items a user clicked in search results.
- Converted Object IDs
Tool to send a conversion event for items outside of search context. use when tracking conversions on category pages or external flows unrelated to algolia search.
- Copy Index
Tool to copy the specified index to a new index. use when you need to duplicate an existing index including records, settings, synonyms, and rules after confirming source and destination names.
- Copy Index Settings
Tool to copy the settings from one index to another. use when you need to replicate index configurations without records or other data.
- Copy Rules
Tool to copy rules from one index to another. use when you need to duplicate all query rules from a source index to a target index after confirming both names.
- Delete Indexdestructive
Tool to delete the specified index and all its records. use when you need to permanently remove an index after confirming it's no longer needed.
- Delete Multiple Recordsdestructive
Tool to delete multiple records from an algolia index. use when you need to remove multiple objects by their ids.
- Delete Ruledestructive
Tool to delete the specified rule from an index. use when you need to permanently remove a rule after confirming its objectid.
- Delete Synonymdestructive
Tool to delete a synonym from a specified index. use when you need to remove an existing synonym by its objectid.
- Export Rules
Tool to export all rules defined on an index. use when you need to backup or migrate index rules.
- Find Object
Tool to find the first object matching a query or filter in an index. use when debugging relevance or filter logic after confirming index exists.
- Get Index Settings
Tool to retrieve the settings of a specified index. use when you need to inspect index configurations after creation or update.
- Get multiple objects
Tool to retrieve multiple records from an index. use when you need to batch-fetch several objectids in one call.
- Get Object Position
Tool to retrieve an object’s position in a result set. use when debugging relevance after performing a search query.
- Index Exists
Tool to check if an algolia index exists. use before performing index operations to prevent accidental index creation. example: indexexists(index name='products').
- Init Insights API Client
Tool to initialize the algolia insights api client. use before sending any insights events.
- List Indices
Tool to list all indices and their metadata. use when you need to retrieve index names, sizes, and state before performing operations that depend on index properties.
- Partial Update Objects
Tool to partially update multiple records in the specified index. use when you need to change only selected fields of many objects without replacing entire records. use after confirming objectids and desired updates.
- Replace All Rules
Tool to push a new set of rules, erasing previous ones. use when you need zero-downtime atomic replacement of all rules in an index.
- Save Synonym
Tool to add or update a synonym in the specified index. use when you need programmatic upsert of search synonyms after index creation.
- Search Algolia Index
Tool to perform a search on a specified algolia index. use after confirming the index name. example: searchindex(index name='contacts', query='apple', search params={'hitsperpage':10})
- Search Multiple Indices
Tool to perform searches across multiple indices in a single call. use when you need to batch multiple index queries into one api request.
- Search Rules
Tool to search for rules in the specified index. use when you need to retrieve rules matching a query, filtering by anchoring, context, pagination, or enabled status.
- Search Synonyms
Tool to search for synonyms in the specified index. use when you need to retrieve synonyms matching a query or filter by type.
- Set Index Settings
Tool to update an algolia index's settings. use when you need to configure index behavior before indexing records. example: set searchableattributes and customranking for products index.
Setup
Setup guide
- 11. In Switchy, open the Space where you want Algolia available and click 'Add Integration' in the sidebar. 2. Select 'Algolia' from the MCP catalog and choose 'API Key' as the auth method. 3. Log into your Algolia dashboard, navigate to Settings > API Keys, and generate a new key with 'addObject', 'deleteObject', 'settings', and 'analytics' scopes (or use an existing Admin API Key if your security policy allows). 4. Paste the API key and your Application ID into Switchy's connection form and click 'Connect'. 5. Test the connection by typing '@Algolia browse the products index' in your Space — if it returns records, you're live. 6. To invoke any tool, @mention Algolia in a message and describe the task in plain language; the AI will map your request to the correct tool and parameters.
What teammates see: by default, memories from Algolia are scoped to the Space (PROJECT visibility) - you can mark any memory PRIVATE or share it ORG-wide.
Works well with
Top models
Compatibility data appears once enough Spaces have used this MCP together with a given model.
How Switchy teams use it
Starter prompts
Browse Index Records
@Algolia browse the products_v2 index and show me the first 20 records with their objectIDs and titlesOpen in a Space →
Clear Staging Index
@Algolia clear all objects from the staging_products index but keep the existing settings and rules intactOpen in a Space →
Clone Production Index
@Algolia copy the products index to products_test so I can experiment with new ranking rules without affecting live searchOpen in a Space →
Create A/B Test
@Algolia create an A/B test on the products index comparing the current config against a variant with typoTolerance set to strict, split traffic 50/50Open in a Space →
Log Conversion Event
@Algolia send a conversion event for objectIDs prod_8472 and prod_9103 from our checkout flow to track which products drive revenueOpen in a Space →
Example outputs
Illustrative - representative of the model's voice and quality, not literal recordings.
@algolia browse the product_catalog index and show me the first 10 records with their objectIDs and titles
Retrieved 10 records from product_catalog index. Sample entries include: objectID '12345' (title: 'Wireless Bluetooth Headphones'), objectID '12346' (title: 'USB-C Charging Cable'), objectID '12347' (title: 'Laptop Stand Aluminum'). The index contains approximately 45,000 total records. Each record includes fields for price, category, inventory_count, and image_url. This represents a snapshot of your current index state.
This example uses the Browse Algolia Index tool to inspect index contents. Requires API key with read permissions on the target index. Useful for auditing data structure before bulk operations, but browsing large indexes (100k+ records) may hit rate limits or require pagination.
@algolia create an AB test comparing our current product_catalog index against product_catalog_v2 to measure click-through rate over the next 14 days
Created AB test 'product_catalog_comparison' with 50/50 traffic split between product_catalog (variant A) and product_catalog_v2 (variant B). Test will run for 14 days starting immediately. Algolia will track click-through rate, conversion rate, and average click position for both variants. You can monitor results in the Algolia dashboard under A/B Tests. Minimum recommended traffic: 1,000 searches per variant for statistical significance.
This example demonstrates the Add AB Test tool for comparing index configurations. Requires write access and an Algolia plan that includes AB testing. The test creates real traffic splits—ensure product_catalog_v2 is production-ready before running. Results typically need 7-10 days to reach significance.
@algolia analyze our search analytics from the past week and identify the top 5 queries with zero results, then suggest which products we should add to fix them
Analyzed search analytics for the past 7 days. Top zero-result queries: 'macbook pro m3' (847 searches), 'wireless earbuds noise cancelling' (612 searches), 'standing desk electric' (438 searches), 'mechanical keyboard rgb' (301 searches), 'usb hub thunderbolt' (289 searches). These represent 2,487 missed opportunities. Recommendation: prioritize adding M3 MacBook Pro models and noise-cancelling earbuds to your catalog, as these account for 58% of zero-result volume and align with your electronics category.
This synthesis example combines Algolia's analytics data with AI reasoning to surface actionable insights. The MCP doesn't have a dedicated 'analyze zero results' tool—the AI would need to query analytics APIs and interpret patterns. Requires analytics API access and sufficient search volume (1,000+ queries/week) for meaningful patterns.
Use-case deep-dives
When Algolia MCP wins for product search optimization
A 6-person e-commerce team running a Shopify store with 8,000 SKUs wants to test whether synonym expansion or query rewriting drives more conversions. The Algolia MCP is the right call here because it exposes the Add AB Test tool directly—you can spin up variant tests comparing index configurations without touching the Algolia dashboard. The team runs a two-week test on their top 200 search queries, tracks click-through with Clicked Object IDs After Search, and logs conversions with Converted Object IDs. The MCP handles all three event types in one workspace, so the PM can review results in Switchy without context-switching. If your catalog is under 10,000 products and you're testing fewer than 5 variants at once, this setup stays fast. Beyond that scale, the API rate limits start to bite and you'll want a dedicated analytics pipeline. For small-team search optimization, this MCP collapses the test-measure-iterate loop into one shared context.
When this MCP simplifies docs index management
A 4-person SaaS startup ships a major docs overhaul every quarter and needs to wipe the old index without breaking search settings or synonyms. The Algolia MCP's Clear Objects tool is built for this—it nukes all records but preserves the index configuration, so the team doesn't have to rebuild rules or replica settings after each reindex. The workflow: the docs lead runs Clear Objects in Switchy, uploads the new markdown corpus via the bulk indexing tool, then spot-checks with Browse Algolia Index to confirm the new records landed. The entire reindex takes 10 minutes instead of the 45-minute dashboard dance they used to do. The trade-off: if your docs corpus is over 50,000 pages, the Browse tool gets sluggish and you'll want to paginate manually. For small-to-mid-size docs sites, this MCP turns quarterly reindexing into a repeatable Switchy workflow that any team member can run.
When Algolia MCP tracks support search behavior
A 5-person support team at a B2B SaaS company wants to know which help articles users click after searching, so they can prioritize content updates. The Algolia MCP's Clicked Object IDs After Search tool logs every click event back to Algolia, and the team can pull click-through reports directly in Switchy without exporting CSVs from the Algolia dashboard. The support lead runs a monthly review: she queries the top 50 search terms, checks which articles got clicked, and flags low-CTR results for rewriting. The MCP also exposes Converted Object IDs, so if the team tracks ticket deflection as a conversion event, they can measure which articles actually solve problems. The limitation: if your knowledge base has more than 20,000 articles or you're logging over 10,000 search events per day, the MCP's event-tracking tools start to lag and you'll need a dedicated analytics stack. For small support teams with focused content libraries, this MCP turns search analytics into a shared team ritual.
Frequently asked
What does the Algolia MCP let me do in Switchy?
It connects your Algolia search indices to Switchy's AI workspace so your team can query records, manage A/B tests, update rules, and send click or conversion events — all without leaving the chat. You can browse entire indices, clear objects, copy indices, and track user interactions directly from prompts.
Do I need admin access to connect Algolia?
You need an Algolia API key with write permissions for most tools — creating A/B tests, clearing objects, and copying indices all require write access. A search-only API key won't work. Check your Algolia dashboard under API Keys to confirm your key's ACL includes addObject, deleteIndex, and editSettings.
Can the Algolia MCP update my index settings or synonyms?
Not directly through the current tool set. You can clear rules, clear objects, and copy an entire index (which includes settings and synonyms), but there's no dedicated tool to patch settings or add synonyms in place. For that, use Algolia's dashboard or REST API alongside this MCP.
Why use this instead of Algolia's dashboard or API?
The MCP lets your team run Algolia operations in plain English inside Switchy — no API docs, no curl commands. It's faster for ad-hoc tasks like checking index contents, running A/B tests, or logging conversion events when you're already collaborating in the workspace. For scheduled jobs or CI/CD, stick with the API.
Who on my team should connect the Algolia MCP?
Whoever owns your search infrastructure — typically a backend engineer or product manager with access to Algolia API keys. Once connected, anyone in the Switchy workspace can invoke the tools, so limit the key's scope if you don't want junior team members clearing production indices.